Not applicable.
Not applicable.
The present invention relates to multislice helical computerized tomography and more particularly to an algorithm, method and apparatus for using the same which reduces the data processing time required to generate an image.
In computerized tomography (CT) x-ray photon rays are directed through a patient toward a detector array. Attenuated rays are detected by the array, the amount of attenuation indicative of the make up (e.g. bone, flesh, air pocket, etc.) of the patient through which the rays traversed. The attenuation data is then backprojected to generate an image of the patient""s internal anatomy.
Early CT systems used a pencil beam photon source consisting essentially of a single ray and a single detector. To collect a complete projection from a single angle about the patient the pencil beam was directed at the patient consecutively from adjacent locations along a line thereby generating parallel ray data for the projection. Other parallel ray projections through the same patient slice from different angles about the slice were generated in the same manner. After multiple (e.g., 500 or more) parallel projection data sets were generated for a slice, the data was backprojected to form a slice image. Because early CT systems generated parallel projection data sets, most CT reconstruction algorithms have been developed assuming parallel data sets.
Unfortunately, while pencil beam systems generate data in a form readily useful with conventional reconstruction algorithms, such systems have a number of shortcomings. One primary shortcoming is that data acquisition periods using such a system are excessive. This is particularly true where images in many slice planes are required. Not only do long acquisition periods reduce system throughput but long periods also often result in relatively poor images. This is because patient movement likelihood increases as the time required for data acquisition increases and patient movement results in blurred images and undesirable artifacts.
Various CT system features and procedures have been developed to increase data acquisition speed including fan beam acquisition, simultaneous multiple slice acquisition and helical scanning. In fan beam systems, instead of using a pencil beam source, the source is collimated into a thin fan beam which is directed at a detector array on a side opposite a patient. In this manner, a complete fan beam projection data set is instantaneously generated for the angle defined by the source during a single data acquisition period and data collection is expedited.
In multiple slice systems, a relatively thick fan beam is collimated and directed at a multi-row detector with a patient therebetween, each detector row in effect gathering data for a separate slice of the thick fan beam along a Z axis perpendicular to the direction of the fan beam.
In a helical scanning system, the source and detector array are mounted on opposing surfaces of an annular gantry and are rotated therearound as a patient is transported at constant speed through the gantry, the x-ray beam sweeps a helical path through the patient, hence the nomenclature xe2x80x9chelical scanning systemxe2x80x9d. Data acquisition can be sped up by increasing the pitch or table translation speed/gantry rotation ratio. Increased pitch typically results in less detailed imaging.
Various combinations of the fan-beam, multislice and helical scanning features have been combined to realize synergies and have been somewhat successful. By combining all three speed enhancing features data acquisition periods are appreciably reduced thereby increasing system throughput and increasing image quality by minimizing the likelihood of patient movement.
While the features described above speed up data acquisition, the resulting data is not in a form which is readily useable with the conventional image reconstruction algorithms. Whereas the conventional algorithms require parallel constant-Z data for reconstruction, data generated using the optimal fast hardware configuration and generation methods generate fan beam (i.e., non-parallel) data for many projections which are not in the same slice (i.e. are multi-Z). Thus, for example, data for two projections will include two separate fan beam projection data sets, a first set at one Z-location and a second set at another Z-location where Z is the axis of gantry rotation.
Not surprisingly, because of data acquisition speed advantages, various algorithms and methods have been developed to generate constant-Z slice images from helical multi-slice fan beam data. To this end, exemplary algorithms require a processor to solve complex and computationally detailed weighting and filtering equations to generate data suitable for backprojection algorithms. Exemplary weighting algorithms are described in an article entitled xe2x80x9cMulti-Slice Helical CT: Scan and Reconstructionxe2x80x9d by Hui Hu which was published in the January 1999 issue of Medical Physics, vol. 26, No. 1, pages 1 through 14. In operation, after imaging data has been collected and archived for a specific patient volume (i.e. 3 dimensional section) of interest, an imaging system operator can select a specific slice and slice thickness through the volume of interest for image reconstruction and display. When a slice is selected, the processor applies the weighting and filtering function to the data to generate the intended image. The weighting function is dependent upon which slice is selected for reconstruction and viewing. Therefore, each time a new slice is selected, a completely different weighting function which is pitch and slice dependent, has to be accessed and applied to the acquired data and the weighted projection data has to be refiltered again to generate the desired image.
Because the filtering and weighting algorithms are extremely complex, data processing is not fast enough to support instantaneous imaging. Thus, after a slice to be imaged is selected, processing requirements cause a delay. The delay is repeated each time a new slice to be imaged is selected. While this process of selection, weighting, evaluation and reselection may not be objectionable where a system user generally knows the slice or slices which should be examined and therefore may only need to be repeated a few times, in some cases the user will not know which images are important and will therefore have to go on a xe2x80x9cfishingxe2x80x9d expedition requiring many iterative image reconstruction sequences. Moreover, where three-dimensional imaging or fluoroscopy techniques are employed most systems require reconstruction of two or more (e.g., some times 6, 12, etc) images per source rotation to generate images having diagnostic quality Z-resolution and temporal resolution. In these cases required reconstruction time is excessive.
Other relatively fast acquisition/processing systems/methods have been developed which include other combinations of fan-beam, multi-slice and helical scanning features. For example, one such system described in an article entitled xe2x80x9cNew Classes of Helical Weighting Algorythms With Applications to Fast CT Reconstructionxe2x80x9d by Guy Besson which was published in Med. Phys. 25(8), August 1998 by Am. Assoc. Phys. Med. combines single slice fan beam data acquisition and helical scanning. As taught in the Besson article such a system is typically used to generate fan beam projections during a single 2 xcfx80 rotation about a patient. Thereafter, the fan beam data is filtered, weighted and backprojected to generate one or more images in various constant Z planes.
Unfortunately, weighting algorithms used with these single slice systems include a fan beam angle dependency and do not lend themselves to fast image reconstruction. This is because, as known in the art, weight distributions present a line of discontinuity across the space of projections which defines two separate sinogram regions. The weighting function expressions differ for the two separate regions. For this reason, reconstruction of P different image planes using a given projection requires P weightings and filterings of that projection.
The Besson article teaches one data processing approach for use with single slice helical fan beam data which reduces processing time appreciably by requiring only one filtering per projection regardless of the number P of image planes required. To this end, the Besson article teaches that single slice fan beam data can be re-binned into multi-Z parallel projections wherein the rays in each parallel projection have the same projection angle. Thus, filtering of the parallel projections need only be performed once to account for the ray parameter.
After filtering, the filtered multi-Z projection data is used to backproject and generate different images within various image planes. Each image is generated as a function of the distance along the Z-axis between a central ray in each filtered multi-Z projection and the plane corresponding to the particular image. In other words, the distances between the rays in each multi-Z projection and the image plane are estimated as being the distance between the central ray in the projection and the image plane.
Unfortunately, the above described system also has a number of shortcomings. Among others, one important shortcoming is that the assumption made during backprojection and helical weighting to generate a planar image introduces an error into the imaging data. This is because, while the estimate is accurate for the central ray in a multi-Z projection, the estimate is less accurate for rays which are positioned laterally in the projection (i.e., each projection consists of multi-Z rays and hence the distance along Z between projection rays and a constant Z image plane is different for each ray).
Another shortcoming is that data acquisition is relatively slow with this single imaging plane architecture (i.e. a single slice detector) when compared to the multi-slice detectors described above.
An exemplary embodiment of the present invention includes a method for use with a CT system which includes a fan beam source and a multi-row detector arranged on opposite sides of a Z-axis wherein the source and detector are rotated about the Z-axis as a patient traverses therealong to generate helical imaging data, the method for generating at least one image within an imaging plane from the helical data. The method comprises the steps of, after fan beam helical imaging data has been acquired, processing the data to generate parallel constant-Z projections proximate the imaging plane, filtering the parallel constant-Z projections and mathematically combining the parallel constant-Z projections as a function of the spatial relationship between the imaging plane and the constant-Z projections to generate at least one image.
Preferably the processing step includes rebinning the fan beam data into parallel multi-Z projections which include parallel rays at different locations and mathematically combining the parallel multi-Z projections to generate the parallel constant-Z projections.
In one embodiment the step of mathematically combining the multi-Z projections includes the step of interpolating between adjacent projections. In another embodiment the step of mathematically combining multi-Z projections includes the step of extrapolating among projections.
In one aspect the step of mathematically combining as a function of the spatial relationships includes the step of mathematically combining as a function of the distances in Z between the imaging plane and the constant-Z projections.
The method is also for generating a second image within a second imaging plane from the helical data and, to this end, comprises the steps of, after filtering, mathematically combining the parallel constant-Z projections as a function of the distance in Z between the second imaging plane and the constant-Z projections to generate the second image.
In another aspect the step of mathematically combining the parallel constant-Z projections includes weighting a sub-set of the constant-Z projections, combining the subset of weighted projections to generate a set of image projections and back-projecting the image projections to generate the image.
Preferably the step of weighting a sub-set includes selecting constant-Z projection pairs, each pair including a first constant-Z projection on a first side of and adjacent the imaging plane and a second constant-Z projection on a second side of and adjacent the imaging plane and weighting each projection pair ray as a function of the distance in Z between the projection including the ray and the image plane and wherein the step of combining the subsets includes, after the projection rays have been weighted, combining the rays in each projection pair into a single projection to be back-projected.
Also, preferably, the method is for generating a second image within a second imaging plane from the helical data and comprising the steps of, after filtering, selecting constant-Z projection pairs, each pair including a first constant-Z projection on a first side of and adjacent the second imaging plane and a second constant-Z projection on a second side of and adjacent the second imaging plane and weighting each projection pair ray as a function of the distance in Z between the projection including the ray and the second image plane, combining the subset of weighted projections to generate a set of image projections and back-projecting the image projections to generate the image.
The invention further includes an apparatus including a processor which runs a pulse sequencing program to perform the methods described above.